Introduction

In plant cells that are enveloped by rigid cell walls, daughter cells are generated within the space enclosed by the mother cell wall. The existence of cell walls probably led to the present diversity of plant cell division methods. For example, unicellular green algae of the Trebouxiophyceae or Chlorophyceae generally multiply via autosporulation, whereas cell division in other green algae such as Spirogyra (Charophyceae) involves the formation of an actin ring followed by the inward growth of a furrow at the cell surface that ultimately leads to the development of a small cell plate at the center of the dividing plane (McIntosh et al. 1995). In comparison, terrestrial plants divide via cell plate formation. The various modes of plant cell division are useful for taxonomy and phylogeny, but they also provide clues concerning the evolution of multicellular higher plants from unicellular green algae.

The class Trebouxiophyceae, which contains the representative genera Chlorella, Parachlorella, and Trebouxia, are non-flagellated vegetative cells that propagate mainly by autosporulation. Nannochloris bacillaris, a long, ellipsoidal, binary-fission-type species first described by Naumann (1921), is also a member of the Trebouxiophyceae (Yamamoto et al. 2001, 2003). In classifying N. bacillaris, Naumann (1921) also included in the genus Nannochloris a small species that divides by binary fission, Nannochloris coccoides. N. bacillaris NRJ propagates by binary fission, whereas N. coccoides CCAP251/1B (Trebouxiophyceae; Yamamoto et al. 2001, 2003) propagates by budding (Tschermak-Woess 1999). Similarly, Marvania geminata, a budding type species identified by Hindák (1976), is also included in the Trebouxiophyceae (Henley et al. 2004). Tschermak-Woess (1999) insisted that N. coccoides CCAP251/1B is really M. geminata and not a Nannochloris species. The Culture Collection of Algae at the University of Göttingen, Germany (SAG), maintains stocks of M. geminata SAG 12.88 and M. geminata SAG 13.96, although the latter is referred to as N. coccoides CCAP251/1B in the Culture Collection of Algae and Protozoa, UK (CCAP).

Recently, species classified into the genus Nannochloris, i.e., N. bacillaris, N. coccoides CCAP251/1B, N. atomus CCAP251/7 and SAG 14.87, N. maculatus, N. eucaryotum, N. oculata, and some unidentified Nannochloris species, were confirmed as Trebouxiophyceae (Yamamoto et al. 2001, 2003; Henley et al. 2004). Phylogenetic analyses based on the actin and 18S rRNA sequences in several Chlorella, Parachlorella, and Trebouxia species indicated that five autospore-forming strains, that is, N. atomus CCAP251/7 and SAG 14.87, N. maculatus, N. eucaryotum, and N. sp. SAG 251-2, are polyphyletic, whereas N. bacillaris and N. coccoides CCAP251/1B are monophyletic and positioned distally (Yamamoto et al. 2003). These results indicate that autosporulation is the ancestral mode of cell division in the Trebouxiophyceae. Non-autosporulative mechanisms such as binary fission and budding would therefore have evolved secondarily. If N. coccoides CCAP251/1B is actually M. geminata, as suggested by Tschermak-Woess (1999), the phylogenetic position of N. coccoides CCAP251/1B and M. geminata should be the same, and both species should be closely related to N. bacillaris. A phylogenetic analysis by Henley et al. (2004) based on existing and new 18S rRNA sequences from multiple Nannochloris species, their related isolates, and numerous other members of the Trebouxiophyceae, grouped N. coccoides CCAP251/1B with M. geminata. The inclusion of numerous related taxa revealed genus-level divergence of N. bacillaris and N. coccoides CCAP251/1B, but both were assigned to the “freshwater group.” Henley et al. (2004) proposed including N. atomus CCAP251/7, N. eucaryotum Mainz1, N. oculata UTEX 1998, N. maculatus CCAP251/3, and some unidentified Nannochloris species in the new genus Picochlorum. According to Henley’s diagnosis, Picochlorum species grow in moist soil or water (either saline or fresh) and Picochlorum is a sister group of the freshwater group in the 18SrRNA tree (Henley et al. 2004). However, N. bacillaris, N. coccoides CCAP251/1B, and M. geminata were assigned to the freshwater group. On the basis of these results, Henley et al. (2004) renamed N. coccoides CCAP251/1B as Marvania coccoides. We reconfirmed the phylogeny of M. coccoides CCAP251/1B and M. geminata SAG 12.88 by constructing a phylogenetic tree based on their actin gene sequences.

By focusing on the relative timing of mother cell wall cleavage and daughter cell wall formation in N. bacillaris and M. geminata SAG 12.88, we sought to understand the relationship between these two species and how their distinct methods of cell division emerged from the ancestral mode of autosporulation. In autosporulation, daughter cells, which have their own walls enveloping their protoplasts, are formed inside the wall of the mother cell. In Chlorella vulgaris, Chlorella sorokiniana, and Chlorella lobophora, which belong to the Trebouxiophyceae, daughter cell wall formation begins at the outer surface of the cell membrane prior to protoplast division. During the division of the protoplast, the daughter cell wall invaginates with the plasma membrane at a cleavage furrow (early synthesis type of daughter cell wall; see Yamamoto et al. 2004). In contrast, in Parachlorella kessleri, which also belongs to the Trebouxiophyceae, synthesis of the daughter cell wall begins at the outer surface of the plasma membrane in the daughter protoplast at the autospore-maturing phase after three successive protoplast divisions (late synthesis type of daughter cell wall; see Yamamoto et al. 2005). In both cases, as the synthesis of the daughter cell wall nears completion, the mother cell wall bursts and the daughter cell emerges with its own wall.

To observe changes in cell wall structure during the cell cycle in N. bacillaris and M. geminata, we used Fluostain I, a cell wall-specific fluorescent dye. We also used transmission electron microscopy with rapid freezing and freeze substitution to observe the structure of the cell walls in detail. We found that the timing of mother cell wall cleavage and exfoliation greatly influences the mode of cell division.

Materials and methods

Algal culture

Nannochloris bacillaris NRJ was collected from Lake Nakatuna in Nagano Prefecture by Dr. N. Takamura of the National Institute for Environmental Studies at Tsukuba, Japan (Ogawa et al. 1995; Arai et al. 1998; Yamamoto et al. 2001) and was given to us by Prof. K. Ueda of Nara Women’s University (NRJ). M. coccoides CCAP251/1B was purchased from CCAP in 1998. M. geminata SAG 12.88 was purchased from SAG in 2002. The species were cultured in modified Ichimura’s C medium (Chida and Ueda 1991) and grown under continuous light (15 μmol photons m−2 s−1) at 23°C.

Light microscopy

Dividing cells were observed under differential interference contrast microscopy (BX52, Olympus, Tokyo, Japan). Cells were harvested by centrifugation at 3,300 × g for 1 min and double-stained with 0.001% Fluostain I (Dojindo, Kumamoto, Japan) and 0.003% SYBR Green I (Molecular Probes, Eugene, OR, USA) in PBS buffer (0.13 M NaCl, 7 mM Na2HPO4, 3 mM NaH2PO4, and pH 7.2). The stained samples were examined first using differential interference and phase-contrast optics for light-field images, and then under ultraviolet excitation for Fluostain I, blue excitation for SYBR Green I, and green excitation for chlorophyll autofluorescence, using a BX52 microscope (Olympus) equipped with a c4742 CCD camera (Hamamatsu Photonics Co., Shizuoka, Japan) and an AQUACOSMOS system (Hamamatsu Photonics Co.). An Olympus Apo 100 ×  objective was used in combination with a U-25ND25 filter and U-MNUA2 band-pass cube.

DNA extraction and actin sequencing

Total DNA was extracted from M. geminata SAG 12.88 and C. lobophora using the Easy-DNA Kit (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. A degenerate PCR method was used to isolate actin gene fragments from M. geminata SAG 12.88 and C. lobophora (Yamamoto et al. 2003). Each PCR product was ligated into the pGEM-T-Easy vector (Promega, Madison, WI, USA). The nucleotide sequences were determined according to Yamazaki et al. (2005), and have been submitted to DDBJ/EMBL/GenBank under accession numbers AB292587 and AB292588 for M. geminata SAG 12.88 and C. lobophora, respectively.

Phylogenetic analysis of the actin gene sequences

The nucleotide sequences of the actin genes from M. geminata SAG 12.88 and C. lobophora were used for phylogenetic analysis, along with the published sequences from N. bacillaris (Arai et al. 1998), M. coccoides CCAP251/1B, N. eucaryotum KSW 0203 (a single colony was isolated from UTEX 2502, Yamamoto et al. 2003), N. atomus SAG 14.87, N. sp. SAG 251-2., Picochlorum atomus CCAP251/7, P. kessleri IAM C-531 (Yamamoto et al. 2001), C. vulgaris IAM C-536, C. sorokiniana IAM C-212, and Trebouxia erici IAM C-593 (Yamamoto et al. 2003). Between 674 and 762 bp of the coding regions from the actin genes in each organism were unambiguously aligned. The first and second positions were used exclusively because of mutational saturation at the third position (Bhattacharya et al. 1991). To test alternative models of evolution for these data, we used PAUP 4.0 beta 10 (Swofford 2002) and Modeltest 3.07 (Posada and Crandall 1998). The tree was constructed using the best fit model (JC + G and γ = 0.1805) selected using hierarchical likelihood ratio testing (hLRT) and the maximum likelihood (ML) algorithm. The robustness of the resulting lineages was tested using bootstrap analysis. The number of bootstrap replicates was 1,000 (minimum evolution) and 1,000 (maximum parsimony). Using the alignment data, a ML analysis was also carried out to estimate quartet puzzling support values with 1,000 puzzling steps. Because Trebouxiophyceae is a sister group of Chlorophyceae within the Chlorophyta (Friedl 1995), the chlorophycean algae Volvox and Chlamydomonas were used as outgroups.

Transmission electron microscopy

Cultured cells of N. bacillaris and M. geminata SAG12.88 were fixed using rapid freeze fixation and freeze substitution and observed using transmission electron microscopy. Rapid freezing was performed according to the method of Baba and Osumi (1987), with minor modifications. Cells were harvested by centrifugation at 3,300 × g for 1 min and then rapidly frozen in liquid propane that had been cooled to −185°C with liquid nitrogen using a Leica EM CPC cryoworkstation (Leica Mikrosystems, Vienna, Austria). The frozen cells were transferred to 2.5% glutaraldehyde in dry acetone at −80°C and incubated at −80°C for 108 h. Subsequently, the samples were warmed gradually from −80 to 0°C over 5 h, held at 0°C for 1 h, warmed from 0 to 23°C over 1 h, and incubated at 23°C for 1 h (Leica EM AFS, Leica Mikrosystems). The samples were washed three times with dry acetone at room temperature and transferred to 2% OsO4 in dry acetone at 40°C for 4 h. The samples were then washed three times with dry acetone at room temperature and infiltrated with increasing concentrations of Spurr’s resin (Spurr 1969) in dry acetone, and finally with 100% Spurr’s resin. Ultrathin sections (0.06–0.07 μm) were cut using a diamond knife on an ultramicrotome (Leica Ultracut UCT, Leica Mikrosystems) and mounted on Formvar-coated copper grids. The sections were stained with 3% uranyl acetate for 2 h at room temperature and with lead citrate for 10 min at room temperature, washed in distilled water, and examined using an electron microscope (H-7600, Hitachi Co., Tokyo, Japan) at 100 kV.

Results

Cell morphology compared to original descriptions of N. bacillaris, M. coccoides, and M. geminata

As described by Naumann (1921), N. bacillaris is ellipsoidal (1.5 μm in width and 3–3.5 μm in length), grows to a length of 5–6 μm, and divides by binary fission (Fig. 1a, d). In comparison, M. coccoides CCAP251/1B is spherical (3–3.2 μm in diameter) and divides by budding (Fig. 1e). Thus, the morphology and mode of cell division observed do not coincide with Naumann’s (1921) description (Fig. 1b). M. geminata SAG 12.88 is spherical (3.5–3.8 μm in diameter) and divides by budding (Fig. 1f), which coincides with Hindák’s (1976) description (Fig. 1c). Our observations indicate that the morphology and mode of cell division in M. coccoides CCAP251/1B match those of M. geminata SAG 12.88.

Fig. 1
figure 1

Modes of cell division in Nannochloris bacillaris, Marvania coccoides CCAP251/1B, and Marvania geminata SAG 12.88. a Nannochloris bacillaris and b Nannochloris coccoides as shown in Naumann’s (19211) original descriptions; they propagate by binary fission. c Marvania geminata as shown in Hindák’s (1976) original description; it propagates by budding. d Differential interference image of N. bacillaris propagating by binary fission. e Differential interference image of M. coccoides CCAP251/1B propagating by budding. f Differential interference image of M. geminata SAG 12.88 propagating by budding. Scale bar 5 μm

Full size image

Phylogenetic analysis using actin gene sequences

We amplified conserved regions of the actin gene in M. geminata SAG 12.88 by degenerate PCR. The sequence of the 674-bp amplified fragment was identical with the sequence reported for M. coccoides CCAP251/1B. The first and second codon positions of the sequences from M. geminata and M. coccoides were used for ML phylogenetic analysis, together with actin gene sequence data from several Nannochloris, Chlorella, and Trebouxia species (Fig. 2). M. geminata SAG 12.88 and M. coccoides CCAP251/1B were positioned at the same site, indicating that they belong to the same genus.

Fig. 2
figure 2

Phylogenetic analysis of the actin-coding regions of Nannochloris and Marvania. A phylogenetic tree was constructed using the maximum likelihood (ML) algorithm in PAUP 4.0 b10. Only the first and second positions of the actin codons were used. Twelve species of Trebouxiophyceae were used as the ingroup; Chlamydomonas reinhardtii and Volvox carteri were used as outgroups. The numbers are bootstrap values (≥50%) based on 1,000 replicates of minimum evolution (ME) and maximum parsimony (MP), and quartet puzzling support values (≥50%) based on ML analyses with 1,000 puzzling steps, respectively. The modes of cell division are illustrated on the phylogenetic tree

Full size image

SYBR Green I/Fluostain I double-staining in cells undergoing binary fission and budding

To compare the process of daughter cell wall synthesis in N. bacillaris and M. geminata SAG 12.88, we observed SYBR Green I/Fluostain I double-stained cells at each stage of the cell cycle. Progress through the cell cycle was judged by SYBR Green I staining of the cell nuclei and the chlorophyll autofluorescence, whereas the synthesis of daughter cell walls was detected by Fluostain I fluorescence.

In N. bacillaris, growth-phase cells were identified by SYBR Green I staining of the single nucleus found at the right side of the image (Figs. 3c, 4c) and autofluorescence of a chloroplast that occupied the lower side of the image (Figs. 3d, 4d). In the early growth phase, strong Fluostain I fluorescence was observed only at the rim on the right side of the image (Fig. 3e). Later in the growth phase, very weak Fluostain I fluorescence was detected throughout the entire cell (Fig. 4e). During protoplast division, the nucleus was situated near the middle of the cell (Fig. 5c). The chloroplast then divided in half (Fig. 5d) and nuclear division followed (Fig. 6c). Strong Fluostain I fluorescence was observed at the plane of division (Fig. 6e). Ultimately, the two daughter cells separated along the plane of division (Figs. 7, 8). No fluorescence was observed in the mother cell wall (Figs. 3e–8e).

Fig. 3-8
figure 3-8

Cell wall development during the cell cycle in Nannochloris bacillaris. Living cells were observed after double-staining with SYBR Green I and Fluostain I. Photographs, from top to bottom, show the progression of the cell cycle. Fig. 3, Early growth phase; Fig. 4, Late growth phase; Fig. 5, Chloroplast division; Fig. 6, Nuclear division; Fig. 7, Protoplast division; Fig. 8, Daughter cell separation. Figs. 3a–8a, Differential interference images; Figs. 3b–8b, Phase contrast images; Figs. 3c–8c, SYBR Green I fluorescence; Figs. 3d–8d, Chlorophyll autofluorescence; Figs. 3e–8e, Fluostain I fluorescence; Figs. 3f–8f, Merged SYBR Green I fluorescence/autofluorescence/Fluostain I fluorescence images. In the false-color images (c–f), DNA is stained green with SYBR Green I, cell walls are stained whitish-blue with Fluostain I, and chlorophyll autofluorescence appears red. Scale bar 5 μm

Full size image

In M. geminata SAG 12.88, cells at the early growth phase were spherical (diameter 3.5–3.8 μm; Fig. 9), with SYBR Green I staining in one nucleus (Fig. 9c) and autofluorescence in one semi-circular chloroplast that occupied half of the cell (Fig. 9d). Fluostain I fluorescence was observed at the edges of each cell. In addition, strong fluorescence was observed at the upper and lower right-hand side of the image (Fig. 9e). Each cell expanded by budding (Figs. 10, 11, 12) from the site of strong Fluostain I fluorescence to the right side of the image (Figs. 10e–12e). The cells grew to 5.8–6.0 μm in length, doubling in volume and assuming a dumbbell shape (Fig. 12). During cell division, the chloroplast divided in half (Fig. 13d), followed by the halving of the nucleus (Fig. 14c). During division of the protoplast, a belt of Fluostain I fluorescence was observed at the budding site (i.e., at the plane of division), which was at the same position as the strong fluorescence observed during the growth phase (Figs. 13e, 14e). Thereafter, the two daughter cells separated (Fig. 15).

Fig. 9-15
figure 9-15

Cell wall development during cell division in Marvania geminata SAG 12.88. Living cells were observed after double-staining with SYBR Green I and Fluostain I. Photographs, from top to bottom, show the progression of the cell cycle. Fig. 9, Early growth phase; Fig. 10, Late growth phase I (the bud is growing); Fig. 11, Late growth phase II (the bud becomes larger); Fig. 12, Chloroplast division; Fig. 13, Nuclear division; Fig. 14, Protoplast division; Fig. 15, Daughter cell separation. Figs. 9a–15a, Differential interference images; Figs. 9b–15b, Phase contrast images; Figs. 9c–15c, SYBR Green I fluorescence; Figs. 9d–15d, Chlorophyll autofluorescence; Figs. 9e–15e, Fluostain I fluorescence; Figs. 9f–15f, Merged SYBR Green I fluorescence/autofluorescence/Fluostain I fluorescence images. In the false-color images (c–f), DNA is stained green with SYBR Green I, cell walls are stained whitish-blue with Fluostain I, and chlorophyll autofluorescence appears red. Scale bar 5 μm

Full size image

Fluorescence of the mother cell wall was not observed in either N. bacillaris or M. geminata SAG 12.88. In contrast, a strong Fluostain I signal at the plane of division was a characteristic feature of both the binary fission and budding type cells.

Daughter cell wall regeneration and mother cell wall exfoliation in N. bacillaris

In autosporulation, the cell sheds its outermost layer (i.e., the mother cell wall) during each round of cell division. We did not observe shedding of the mother cell wall in either binary fission or budding. Further, the strong Fluostain I signal observed at the plane of division was common to both species, suggesting that active daughter cell wall synthesis occurs there. Based on these results, we used electron microscopy in combination with rapid freeze fixation and freeze substitution to explore daughter cell wall regeneration and mother cell wall exfoliation.

During the early growth phase of N. bacillaris, each cell was ellipsoidal in shape with a large ellipsoidal chloroplast and a nucleus positioned at one end of the cell (Fig. 16a). Later in the growth phase, the cell elongated with the chloroplast (Fig. 16b), and a cell wall was observed (cw1, the outer electron-dense layer; Fig. 16e, f). During protoplast division (Fig. 16c), invagination of the plasma membrane was observed at the plane of division (Fig. 16g), and synthesis of the daughter cell wall on the plasma membrane had not yet begun (Fig. 16d, i). This suggests that the daughter cell wall is not synthesized by the time the protoplast completes division (Fig. 16j).

Fig. 16
figure 16

Transmission electron micrographs of Nannochloris bacillaris in the growth phase (a, b, e, and f) and during protoplast division (c, d, and g–j). a A whole cell in the early growth phase, b a whole cell in the late growth phase, c a whole cell showing invagination of the plasma membrane, d a whole cell after invagination of the plasma membrane, e highly magnified view of the cell wall in a, f highly magnified view of the cell wall in b, g highly magnified view of the plane of division in c, h the site of invagination of the plasma membrane at the cell surface in g, i highly magnified view of the plane of division in d, j the site of invagination of the plasma membrane at the cell surface in i. Scale bar 1 μm (d) and 100 nm (i and j)

Full size image

Instead, shortly after protoplast division, a thin daughter cell wall (i.e., cw2) was synthesized between the invaginated plasma membranes at the plane of division (Fig. 17a, e, f) and gradually increased in thickness (Fig. 17b, g, h). A triangular space was observed between cw1 and cw2 (Fig. 17f, h, arrows); the space was smaller than that observed during autosporulation, so it could not be observed under light microscopy. Because the daughter cell wall (cw2) was newly synthesized, cw1 is hereafter called the mother cell wall. The mother cell wall was cleaved at the triangular space (Fig. 17c) on the upper side of the image (Fig. 17i), whereas the lower side remained intact (Fig. 17j).

Fig. 17
figure 17

Transmission electron micrographs of Nannochloris bacillaris during formation of the daughter cell wall (a, b, and e–h) and during separation of the daughter cells (c, d, and i–k). a A whole cell during early daughter cell wall formation, b a whole cell during late daughter cell wall formation, c a whole cell showing cleavage of the mother cell wall at the plane of division, d a whole cell showing the separation of the daughter cells, e highly magnified view of the cell wall at the plane of division in a, f the plane of division at the cell surface in e, g highly magnified view of the cell wall at the plane of division in b, h the plane of division at the cell surface in g, i highly magnified image of c at the upper side of the plane of division, j highly magnified image of c at the lower side of the plane of division, k highly magnified view of d at the plane of daughter cell separation. Scale bar 1 μm (d) and 100 nm (h and k). White arrows, triangular space; black arrows, cleavage site of cw1

Full size image

After cleavage of the mother cell wall, the daughter cell divided into two from the upper side of the plane of division (Fig. 17d). The plane of division then separated, exposing cw2 to the outside (Fig. 17k). Strong Fluostain I signals were detected at these positions (Fig. 3e). However, cw1 did not separate from the daughter cell; instead, it remained adhered to cw2. A scar from the cleavage of cw1 remained on the surface of the daughter cell just after cell division (Fig. 17k, arrow). The cleaved mother cell wall degenerated gradually and was not observed in the next growth phase (Fig. 16e, f).

Daughter cell wall regeneration and multiple mother cell wall layers in M. geminata

Early in the growth phase, M. geminata was spherical (Fig. 18a) with a multilayered cell wall. The cw1 enclosed the entire cell and was overlaid by the grandmother cell wall (cw0); however, cw0 did not completely enclose the entire cell (Fig. 18e). Breaks were detected that seemed to coincide with strong Fluostain I fluorescence (see Fig. 9e). In the budding cell, the bud was not covered by cw0 (Fig. 18b). Instead, the cell grew from a site that was free from remnants of the cleaved cw0 (Fig. 18f). Following protoplast division, cw2 gradually increased in thickness between the invaginated plasma membrane (Fig. 18c). The cw1 is hereafter called the mother cell wall (Fig. 18g). Although the triangular space between cw1 and cw2 was not obvious, cw1 was cleaved at that position and the daughter cells separated at the plane of division (Fig. 18d). Consequently, the daughter cell on the left side of the image was covered by cw2, cw1, and cw0; the daughter cell on the right side was covered by cw2 and cw1 (Fig. 18h). The two daughter cells were consistently asymmetrical in terms of their cell wall formation. The mother cell wall was not exfoliated as in autosporulation, and it did not degenerate as in N. bacillaris. The daughter cells entered the next growth phase still carrying the mother cell wall or grandmother cell wall, causing the uncovered portion of the cell to bud outward.

Fig. 18
figure 18

Transmission electron micrographs of Marvania geminata SAG 12.88 during the growth phase (a, b, e, and f) and during cell division (c, d, g and h). a A whole cell in the early growth phase, b a whole cell in the bud-forming phase, c a whole cell during formation of the daughter cell wall after protoplast division, d a whole cell during separation of the daughter cells after cleavage of the mother cell wall, e and f highly magnified views of the cell walls in a and b at the budding site, g and h highly magnified views of the cell walls in c and d at the plane of division. Scale bar 1 μm (d) and 100 nm (h)

Full size image

Discussion

We reconfirmed that in M. coccoides CCAP251/1B (Fig. 1e), cell division does not occur as described by Naumann (1921; Fig. 1b). In addition, the results of our phylogenetic analysis using actin sequence data from several species, including M. geminata SAG 12.88, indicate that M. coccoides CCAP251/1B and M. geminata SAG 12.88 are located in the same position. Tschermak-Woess (1999) insisted that N. coccoides CCAP251/1B (i.e., M. coccoides) is not a Nannochloris species, but that it is actually M. geminata (Hindák 1976; Fig. 1c), which reproduces by budding. The 18S rRNA sequence data from M. geminata SAG 12.88 was submitted to DDBJ/EMBL/GenBank by Bhattacharya et al. (1999; accession number AF124336). Henley et al. (2004) used 18S rRNA sequence data to construct a detailed phylogenetic tree that included M. geminata SAG 12.88 and a new isolate, M. sp. JL11-11. Our results indicate that these species are all close relatives. There are differences in cell size, cultured cell buoyancy, and extracellular polysaccharide distribution between M. coccoides CCAP251/1B and M. geminata SAG12.88. It is thus appropriate that Henley et al. (2004) renamed N. coccoides CCAP251/1B as M. coccoides. Among the strains presently available, N. bacillaris is the only “true” Nannochloris species that matches Naumann’s (1921) original description (i.e., is ellipsoidal in shape and propagates by binary fission).

Henley et al. (2004) proposed including N. atomus CCAP251/7 in the new genus Picochlorum, with some other Nannochloris species. However, N. atomus SAG 14.87 and N. eucaryotum UTEX 2502 were excluded from Picochlorum because of markedly different 18SrRNA sequences. However, in this study, the actin tree suggests a close relationship among these three species, P. atomus CCAP251/7, N. atomus SAG 14.87, and N. eucaryotum UTEX 2502 (Fig. 2). The incongruity between the actin and 18SrRNA trees was discussed in Yamamoto et al. (2003).

The stilbene-derived fluorescent dye Fluostain I has been used extensively to study plant morphogenesis (Belliveau et al. 1990). It is possible to visualize the cell walls of live Trebouxiophyceae by fluorescent microscopy using Fluostain I. In the case of autosporulation, one can observe both the fluorescence of the daughter cell wall, as well as the sloughing of the mother cell wall by staining cells in the hatching phase (Yamamoto et al. 2004). Shedding of the mother cell wall is relatively difficult to observe by transmitted light, but it is easily observed by Fluostain I staining. However, we observed no sloughing of the mother cell wall in N. bacillaris and M. geminata. Shedding of the mother cell wall is not a common feature of binary fission and budding. The primary difference between binary fission/budding and autosporulation is whether the mother cell wall is shed; however, there are other differences between binary fission and budding. For example, in M. geminata, a budding type alga, we were unable to differentiate the mother cell wall from the daughter cell wall; a fluorescent signal was observed across the entire cell. In contrast, in N. bacillaris, the newly formed cell wall at the plane of division displayed a stronger signal than any other area in the cell. Moreover, we observed spots of strong fluorescence at the growth phase during budding that were never observed during binary fission.

Sluiman and Reymond (1987) observed mitosis and cytokinesis in chemically fixed M. geminata using electron microscopy. They demonstrated that after the formation of the new wall layers at the septum, the mother cell wall that encloses the two daughter cells breaks at the position of the septum. They reported that the pattern of cell division in M. geminata appeared to be more similar to that of the Chlorococcales than suggested by initial light microscopic observations (Hindák 1976). They also noted that a better description of cell division in M. geminata may be “semi-exogenous formation of autospore diads” (Sluiman and Reymond 1987).

We observed N. bacillaris and M. geminata SAG 12.88 using electron microscopy in combination with rapid freeze fixation and freeze substitution. Formation of the daughter cell wall and cleavage of the mother cell wall occurred at the plane of division, and the two daughter cells separated. This sequence of events corresponds to the term “semi-exogenous formation of autospore diads.” Because the daughter cell remains covered by the remnants of the cleaved mother cell wall, the growth of the daughter cell proceeds by expansion from a site not covered by mother cell wall. This results in bulging or budding of the growing cell. It has been suggested that the essential processes of cell division (i.e., protoplast division, daughter cell wall synthesis, and cleavage of the mother cell wall) are the same as in autosporulation. Indeed, some members of the genus Marvania, which includes the subaerial tropical green alga M. aerophytica (Neustupa and Šejnohová 2003), reproduce via a type of budding that has been described as highly modified autosporulation.

Our phylogenetic analysis and transmission electron microscopy observations indicate that non-autosporic cell division arose from the autosporic mode due to the loss of shedding of the mother cell wall. This process likely involved a series of evolutionary events. For example, a mode of cell division that did not involve hatching of the daughter cell may have been followed by a mode in which the mother cell wall gradually degenerated during the subsequent growth phase (binary fission), until finally, a mode emerged in which the cleaved mother cell wall remained covering the daughter cell during the next growth phase (budding; Fig.19). An essential difference between autosporulation and non-autosporulative binary fission and budding is whether the mother cell wall is shed. In addition, a difference in the onset of mother cell wall degeneration distinguishes binary fission from budding. During binary fission in N. bacillaris, the daughter cell wall does not form during the invagination of the plasma membrane at the plane of division (Fig. 16h, j). Similarly, in M. geminata, invagination of the plasma membrane alone was observed at the plane of division during budding (data not shown). These results suggest that binary fission and budding share a common ancestral non-hatching mode of cell division of the daughter cell wall-late synthesis type. Many ellipsoidal species propagate by binary fission. There are other ellipsoidal trebouxiophyceaen genera; for example, Stichococcus (Pickett-Heaps 1974; Neustupa et al. 2007) and Koliellopsis (Lokhorst et al. 2004), in which the cells unite to form filaments. In non-hatching N. bacillaris, if cleavage of the mother cell wall or separation of the daughter cells does not occur, the cells will join together to form filaments. The parallel emergence of multicellular algal species may have begun from processes such as filament formation.

Fig. 19
figure 19

Evolution of binary fission and budding from autosporulation. This scheme summarizes the timing of mother cell wall (MCW) cleavage and daughter cell wall (DCW) formation in three modes of cell division: autosporulation (modified from Yamamoto et al. 2004, 2005), binary fission, and budding. The divergence of early type daughter cell wall synthesis in autosporulation, late-type daughter cell wall synthesis in autosporulation, binary fission, and budding is indicated according to the topography of the phylogenetic tree based on actin sequences (left). A lack of shedding of the mother cell wall may have led to the development of binary fission and budding

Full size image